WillFlame14 / hanabi-bot

A bot that plays on the hanab.live interface.
GNU General Public License v3.0
14 stars 9 forks source link

Incorrect finesse inferrence based on uncertain knowledge #232

Open flackr opened 1 month ago

flackr commented 1 month ago

Version (PM the bot with /version): v1.4.2 Convention settings: /setall 11 Steps to reproduce or replay link: https://hanab.live/replay/1169822#11

Additional information: On turn 11, when robot1 tells robot2 about 4's, it can't be r4 because robot1 doesn't yet know whether their red is r2 or r3 until botbot1 has a chance to play. robot2 incorrectly has robot1 as knowing that it is r3 based on the 4 clue.

WillFlame14 commented 3 weeks ago

I'm not sure the best way to handle this. robot2 assumes that robot3 knows about their own r3 asymmetrically, since this clue has been given before if robot1 is sure that a finesse is occurring (e.g. if robot2 has the second copy of r2 in their hand). I'm not sure whether to disallow that type of asymmetric clue or to disallow the 4 clue from robot1. I've definitely seen the asymmetric clue given a few times, and expecting robot1 to double self-finesse in this situation seems very complicated compared to just r4, but the team can't be fully synced on this. :(